Data-Driven Adaptive Compensation of Tool Runout in Milling via Machine Tool Feed Drives

Abstract

Abstract. With recent advances in computer numerical control (CNC) systems, modern machine tools have become increasingly intelligent, capable of automatically compensating for process errors and abnormalities. A well-known source of errors in modern milling processes is associated with tool eccentricity and cutter runout that produce rough surface finish and lead to accelerated tool wear. This article presents a novel strategy where the CNC machine tool senses the eccentricity/runout related errors on-the-fly and compensates for them using its own feed drive system. A general formulation is developed, which reveals the force/vibration frequency spectrum of the milling process suffering from tool eccentricity/radial runout. The tool eccentricity is then compensated by commanding the machine tool feed drives with microcircular trajectory at the spindle frequency, which then cancels the circular (eccentric) motion of the tool center point. The commanded circular trajectory parameters, i.e., the amplitude and the phase, are adjusted automatically by iteratively learning the dynamic response of the feed drive system and using the tool eccentricity-induced process response based on the data collected either via an accelerometer or a force sensor. The overall learning (adaptation) process is formulated as a convex optimization problem, and various simulation studies are provided to demonstrate the optimality and the convergence of the approach. The effectiveness of the proposed strategy is validated through various simulation studies and actual milling experiments.